Some time ago, I posted my question concerning
devfs/devfsd in 2.4.2 registering "generic" name offered by
sg module while "cd" offered by sr_mod.o was not registered
for my SCSI CDs.
Namely, I needed to load "sr_mod.o" somehow on my own
while "sg" seemed to be loaded for no apparent reason.
Chiaki,
Wow! This is proving a very time consuming
problem. There is one small question I can
answer:
Ishikawa wrote:
[snip]
subtle differences the way devfs is handled
Why does sg.c use write_lock_irqsave() while sr.c does not?
The lock is there because st and sg have got rid of
the
I was trying to figoure out "WHY" sg was loaded
since I was not asking for it and /etc/init.d/* scripts
seemed not to.
I would think a prudent first step in tracking this down would be to chop
out parts of the init processing until sg.o no longer gets loaded at
initialization time, and thus
Doug,
Thank you for the enlightment about write_lock_irqsave().
Regarding who is doing sg loading,
I will try your suggestions over the weekend to
find out.
Knowing what the current task and its
parent task seem to be a great help
in analyzing the system behavior and learning aid.
(I wonder
4 matches
Mail list logo